tasks: coci doi only; COCIDOIOnly

1 files changed, 18 insertions, 2 deletions

diff --git a/python/refcat/techreport.py b/python/refcat/techreport.py
index 904f9c9..064e357 100644
--- a/python/refcat/techreport.py
+++ b/python/refcat/techreport.py
@@ -4,11 +4,27 @@ Tasks for techreport.
 """
 
 from refcat.tasks import Refcat, OpenCitations
+from refcat.base import shellout, Zstd
 
-class COCIStats(Refcat):
+
+class COCIDOIOnly(Refcat):
+    """
+    Extract DOI-DOI pair, order dois lexicographically.
+    """
 
     def requires(self):
         return OpenCitations()
 
     def run(self):
-        print(self.input().path)
+        """ https://unix.stackexchange.com/a/37470/376 """
+        output = shellout("""
+                          zstdcat -T0 {input} |
+                          tail -n +2 |
+                          cut -d , -f2,3 |
+                          perl -F, -lane 'printf qq[%s\n], join ",", sort @F' |
+                          zstd -c -T0 > {output}
+                          """, input=self.input().path)
+        luigi.LocalTarget(output).move(self.output().path)
+
+    def output(self):
+        return luigi.LocalTarget(path=self.path(ext="tsv.zst"), format=Zstd)